DP83822H: MDIO Serial Interface Issues

Hi Luigi

Can you read registers 2 and 3 and let me know their values? Also can you share your schematic? 

Thank you,

Nikhil

Hi Nikhil,

Reg 2 = 0x2000;

Reg 3 = 0xA240;

I'm sorry, but I can't share the schematic. If you have any doubt, I can check.

Thanks

Luigi

Hi Luigi,

Please fill out the schematic checklist from this link: https://sps16.itg.ti.com/sites/HSDCOVERVIEW/Ethernet/_layouts/15/WopiFrame.aspx?sourcedoc={521B8C32-FDD4-4216-9D33-7A8DF4A2384D}&file=DP83822_%20Schm_Checklist_StrapTool.xlsx&action=default

Thank you,

Nikhil

Hi Nikhil,

sorry for the delay. We were closed for the Christmas holidays. The problem still exists. I tried to open your link but it doesn't work.

Hi Luigi,

I have attached the schematic checklist, please let me know if you have any concerns. 

Thank you,

Nikhil

Hi Nikhil,

We will check the schematic the next week. Now I can't.

At the moment I can show you one read request on address 0X0019

How you can see, every time before to send a write request or a read request there is a 32 MDC clock cycle pause, then (in this case) there is the read request and the following answer. In this case I'm polling on this register.

Regards

Luigi 

Hi Luigi,

Can you clarify what is the pause you are seeing? Is it between sending a read command and reading the actual value back? Or is it between subsequent read commands? You initially were having issues with writing registers. Can you send me a snapshot of the clock cycles of reading a register with the PHY default setting, then a snapshot of writing to that register, and then a snapshot of reading that register back?

Please let me know when you have completed the schematic review, and let me know if you see any issues.

Thank you,

Nikhil 

Hi Nikhil,

The schematic seems to be ok.

The pause is between subsequent read commands. My snapshoot refeers to this issue:

- When I try to get the MII Link Status (Bit[11]) in reg 0x0019, I will be able to detect the link status in Bit[10] of reg 0x0019;

Do you think that your request could help? I'm able to read and write some registers and not others. For example in reg 0x0019 I got the previous issue

Hi Luigi,

We want to track the write cycle and the read cycle before and after the write command was made to make sure the SMI protocol is getting clocked correctly. Table 7 in the datasheet shows the SMI protocol. This will help us investigate why a read command may not set a bit correctly. 

Thank you,

Nikhil 

Hi Nikhil,

these snapshots refers to 

write in 0x19 0x0080

read 0x19 (link off)

read 0x19 (link on)

write in 0x19 0x0000

read in 0x19 (link off)

read in 0x19 (link on)

Hi Luigi,

Nikhil is out of office so I can step in here to help. If I understand your problem correctly, you are repeatedly trying to read register 0x19 and see incorrect values. Can you tell me if the register read value is correct the first time that you read it?

What is the pull up resistor value on MDIO pin? Is it pulled up to VDDIO voltage? Are there other PHYs on the same MDIO-MDC bus?

What is the MDC frequency in your application? If its at the higher end (25MHz), can you try with slower MDC frequency?

Lastly, can you try adding more clock cycles between subsequent read i.e. instead of 32 cycles try with 34 cycles.

-Regards

Aniruddha

Hi Aniruddha,

Aniruddha Khadye said:

Nikhil is out of office so I can step in here to help. If I understand your problem correctly, you are repeatedly trying to read register 0x19 and see incorrect values. Can you tell me if the register read value is correct the first time that you read it?

It never works.

Aniruddha Khadye said:

What is the pull up resistor value on MDIO pin? Is it pulled up to VDDIO voltage? Are there other PHYs on the same MDIO-MDC bus?

2.5V and pullup 2.2K 

Aniruddha Khadye said:

What is the MDC frequency in your application? If its at the higher end (25MHz), can you try with slower MDC frequency?

I tried 1 and 2 Mhz

Aniruddha Khadye said:

Lastly, can you try adding more clock cycles between subsequent read i.e. instead of 32 cycles try with 34 cycles.

I tried but it still doesn't work

Hi Luigi,

Are you using multiple PHYs in the design? Are there multiple pull-ups along the MDIO line?

Would you be able to provide the schematic over private message or email?

Additionally, you mentioned you are able to read register 1 correctly. Can you read back registers 0 to 1F multiple times, and see which ones you are able to read back correctly?

Thank you,

Nikhil 

Hi Nikhil,

At address 00h the value is 2100h

At address 01h the value is 7849h

And next ones

I'm going to send in mp the schematic

Hi Luigi,

We are looking into this issue and will provide an update Wednesday.

Thank you,

Nikhil

Hi Luigi,

The register read information looks ok. Are you able to successfully read and write registers on other PHYs? Additionally, could you try accessing the registers using our USB-2-MDIO tool?

Thank you,

Nikhil

Hi Nikhil,

How do explain this? 

Luigi Pirillo said:

these snapshots refers to 

write in 0x19 0x0080

read 0x19 (link off)

read 0x19 (link on)

write in 0x19 0x0000

read in 0x19 (link off)

read in 0x19 (link on)

Hi Luigi,

The op-code bits look good from the screen captures provided.

What is the processor IO voltage? The PHY IO voltage is 2.5 V and the processor voltage much match. Additionally, are the two MDIO lines tied together? If so, you will only need a single 2.2k pull-up resistor. The second one can be removed. 

Just as a sanity check, are you able to correctly write registers of other devices? Are you able to communicte with our USB-2-MDIO tool?